: Propensity score (PS) analysis is increasingly being used in observational studies, especially in some cancer studies where random assignment is not feasible. This systematic review evaluates the ...use and reporting quality of PS analysis in oncology studies.
: We searched PubMed to identify the use of PS methods in cancer studies (CS) and cancer surgical studies (CSS) in major medical, cancer, and surgical journals over time and critically evaluated 33 CS published in top medical and cancer journals in 2014 and 2015 and 306 CSS published up to November 26, 2015, without earlier date limits. The quality of reporting in PS analysis was evaluated. It was also compared over time and among journals with differing impact factors. All statistical tests were two-sided.
More than 50% of the publications with PS analysis from the past decade occurred within the past two years. Of the studies critically evaluated, a considerable proportion did not clearly provide the variables used to estimate PS (CS 12.1%, CSS 8.8%), incorrectly included non baseline variables (CS 3.4%, CSS 9.3%), neglected the comparison of baseline characteristics (CS 21.9%, CSS 15.6%), or did not report the matching algorithm utilized (CS 19.0%, CSS 36.1%). In CSS, the reporting of the matching algorithm improved in 2014 and 2015 ( P = .04), and the reporting of variables used to estimate PS was better in top surgery journals ( P = .008). However, there were no statistically significant differences for the inclusion of non baseline variables and reporting of comparability of baseline characteristics.
The use of PS in cancer studies has dramatically increased recently, but there is substantial room for improvement in the quality of reporting even in top journals. Herein we have proposed reporting guidelines for PS analyses that are broadly applicable to different areas of medical research that will allow better evaluation and comparison across studies applying this approach.
Cancer outcomes vary widely among different countries. However, comparisons of cost-effectiveness and cost-efficiency of different systems are complex because the incidences of different cancers vary ...across countries and their chances of cure also differ substantially. We aim to propose a new standardized method for global comparison and to explore its relationship with economic indicators.
Cancer statistics from all 184 countries and 27 cancers listed in GLOBOCAN 2012 were analyzed. The complement of age-standardized mortality/incidence ratio 1 - (ASM/ASI) was taken as the proxy relative survival (RS). Accounting for various country-specific cancer patterns, the cancer site-standardized proxy RS (proxy SS-RS) of individual countries were calculated by weighting the proportion of specific cancer sites as compared with the global pattern of incidence. Economic indicators of different countries listed by the World Bank were correlated with corresponding proxy SS-RS.
Substantial variation in site-specific survival and new case distribution supported the use of proxy SS-RS, which ranged from 0.124 to 0.622 (median 0.359). The median total health expenditure per capita (HEpc) increased from US$44 for countries with proxy SS-RS < 0.25, to US$4643 for countries with proxy SS-RS ≥0.55. Results from logarithmic regression model showed exponential increase in total HEpc for better outcome. The expenditure varied widely among different strata, with the widest difference observed among countries with SS-RS ≥0.55 (total HEpc US$1412-$9361).
Similar to age-standardization, cancer site-standardization adjusted for variation in pattern of cancer incidence provides the best available and feasible strategies for comparing cancer survivals across countries globally. Furthermore, cancer outcome correlated significantly with economic indicators and the amount of HEpc escalated exponentially. Our findings call for more in-depth studies applying cancer-site standardization to provide essential data for sharing of experience and urgent actions by policy makers to develop comprehensive and financially sustainable cancer plan for greater equity.
Background The short-term superiority of video-assisted thoracoscopic surgery lobectomy compared with open lobectomy for early-stage lung cancer has been suggested by single-institution studies. Lack ...of equipoise limits the feasibility of a randomized study to confirm this. The hypothesis of this study (CALGB 31001) was that VATS lobectomy results in shorter length of hospital stay and fewer complications compared with open lobectomy in stages I and II non-small cell lung cancer in a multi-institutional setting. Methods Five hundred nineteen patients whose tumors had been collected as part of CALGB 140202 (lung cancer tissue bank) were eligible. Propensity-scoring using age, race, sex, performance status, comorbidities, histology, tumor stage, and size as independent variables was used to create a 1:1 matched group of 175 pairs of patients. McNemar’s test for binary variables and Wilcoxon signed-rank test for continuous variables were used to assess differences in length of hospital stay, complications, and discharge dispositions between the groups. Comparison of disease-free and overall survival between the two approaches was done using the log-rank test. Probability values of less than 0.05 were considered significant. Results The matched data on length of hospital stay, complications, and discharge dispositions significantly favored the video-assisted thoracoscopic surgery group. There was no statistically significant difference in survival between the two approaches. Conclusions This multi-institutional study supports the assertion that thoracoscopic lobectomy results in shorter hospital length of stay, fewer perioperative complications, and greater likelihood of independent home discharge compared with open lobectomy for early-stage lung cancer. Survival was comparable between the two groups.
IMPORTANCE: Patients with schizophrenia have a substantially higher suicide rate than the general public. Early intervention (EI) services improve short-term outcomes. However, little is known about ...the association of EI with suicide reduction in the long term. OBJECTIVE: To examine the association of a 2-year EI service with suicide reduction in patients with first-episode schizophrenia-spectrum (FES) disorders during 12 years and the risk factors for early and late suicide. DESIGN, SETTING, AND PARTICIPANTS: This historical control study compared 617 consecutive patients with FES who received the 2-year EI service in Hong Kong between July 1, 2001, and June 30, 2003, with 617 patients with FES who received standard care (SC) between July 1, 1998, and June 30, 2001, matched individually. Clinical information was systematically retrieved for the first 3 years of clinical care for both groups. The details of death were collected up to 12 years from presentation to the services. Data analysis was performed from October 30, 2016, to August 18, 2017. MAIN OUTCOMES AND MEASURES: Suicide rate during 12 years was the primary measure. The association of the EI service with the suicide rates during years 1 through 3 and years 4 through 12 were explored separately. RESULTS: The main analysis included 1234 patients, with 617 in each group (mean SD age at baseline, 21.2 3.4 years in the EI group and 21.3 3.4 years in the SC group; 318 male 51.5% in the EI group and 322 52.2% in the SC group). The suicide rates were 7.5% in the SC group and 4.4% in the EI group (McNemar χ2 = 5.55, P = .02). Patients in the EI group had significantly better survival (propensity score–adjusted hazard ratio, 0.57; 95% CI, 0.36-0.91; P = .02), with the maximum association observed in the first 3 years. The number of suicide attempts was an indicator of early suicide (1-3 years). Premorbid occupational impairment, number of relapses, and poor adherence during the initial 3 years were indicators of late suicide (4-12 years). CONCLUSIONS AND RELEVANCE: This study suggests that the EI service may be associated with reductions in the long-term suicide rate. Suicide at different stages of schizophrenia was associated with unique risk factors, highlighting the importance of a phase-specific service.
High-throughput sequencing data are widely collected and analyzed in the study of complex diseases in quest of improving human health. Well-studied algorithms mostly deal with single data source, and ...cannot fully utilize the potential of these multi-omics data sources. In order to provide a holistic understanding of human health and diseases, it is necessary to integrate multiple data sources. Several algorithms have been proposed so far, however, a comprehensive comparison of data integration algorithms for classification of binary traits is currently lacking.
In this paper, we focus on two common classes of integration algorithms, graph-based that depict relationships with subjects denoted by nodes and relationships denoted by edges, and kernel-based that can generate a classifier in feature space. Our paper provides a comprehensive comparison of their performance in terms of various measurements of classification accuracy and computation time. Seven different integration algorithms, including graph-based semi-supervised learning, graph sharpening integration, composite association network, Bayesian network, semi-definite programming-support vector machine (SDP-SVM), relevance vector machine (RVM) and Ada-boost relevance vector machine are compared and evaluated with hypertension and two cancer data sets in our study. In general, kernel-based algorithms create more complex models and require longer computation time, but they tend to perform better than graph-based algorithms. The performance of graph-based algorithms has the advantage of being faster computationally.
The empirical results demonstrate that composite association network, relevance vector machine, and Ada-boost RVM are the better performers. We provide recommendations on how to choose an appropriate algorithm for integrating data from multiple sources.
To make a systematic review of risk factors, outcomes and prevalence of extended-spectrum β-lactamase-associated infection in children and young adults in South-East Asia and the Western Pacific.
Up ...to June 2018 we searched online databases for published studies of infection with extended-spectrum β-lactamase-producing Enterobacteriaceae in individuals aged 0-21 years. We included case-control, cohort, cross-sectional and observational studies reporting patients positive and negative for these organisms. For the meta-analysis we used random-effects modelling of risk factors and outcomes for infection, and meta-regression for analysis of subgroups. We mapped the prevalence of these infections in 20 countries and areas using available surveillance data.
Of 6665 articles scanned, we included 40 studies from 11 countries and areas in the meta-analysis. The pooled studies included 2411 samples testing positive and 2874 negative. A higher risk of infection with extended-spectrum β-lactamase-producing bacteria was associated with previous hospital care, notably intensive care unit stays (pooled odds ratio, OR: 6.5; 95% confidence interval, CI: 3.04 to 13.73); antibiotic exposure (OR: 4.8; 95% CI: 2.25 to 10.27); and certain co-existing conditions. Empirical antibiotic therapy was protective against infection (OR: 0.29; 95% CI: 0.11 to 0.79). Infected patients had longer hospital stays (26 days; 95% CI: 12.81 to 38.89) and higher risk of death (OR: 3.2; 95% CI: 1.82 to 5.80). The population prevalence of infection was high in these regions and surveillance data for children were scarce.
Antibiotic stewardship policies to prevent infection and encourage appropriate treatment are needed in South-East Asia and the Western Pacific.
Purpose
Biomedical data frequently contain imbalance characteristics which make achieving good predictive performance with data-driven machine learning approaches a challenging task. In this study, ...we investigated the impact of re-sampling techniques for imbalanced datasets in PET radiomics-based prognostication model in head and neck (HNC) cancer patients.
Methods
Radiomics analysis was performed in two cohorts of patients, including 166 patients newly diagnosed with nasopharyngeal carcinoma (NPC) in our centre and 182 HNC patients from open database. Conventional PET parameters and robust radiomics features were extracted for correlation analysis of the overall survival (OS) and disease progression-free survival (DFS). We investigated a cross-combination of 10 re-sampling methods (oversampling, undersampling, and hybrid sampling) with 4 machine learning classifiers for survival prediction. Diagnostic performance was assessed in hold-out test sets. Statistical differences were analysed using Monte Carlo cross-validations by post hoc Nemenyi analysis.
Results
Oversampling techniques like ADASYN and SMOTE could improve prediction performance in terms of G-mean and F-measures in minority class, without significant loss of F-measures in majority class. We identified optimal PET radiomics-based prediction model of OS (AUC of 0.82, G-mean of 0.77) for our NPC cohort. Similar findings that oversampling techniques improved the prediction performance were seen when this was tested on an external dataset indicating generalisability.
Conclusion
Our study showed a significant positive impact on the prediction performance in imbalanced datasets by applying re-sampling techniques. We have created an open-source solution for automated calculations and comparisons of multiple re-sampling techniques and machine learning classifiers for easy replication in future studies.
To evaluate the efficacy of maintenance sunitinib after chemotherapy for small-cell lung cancer (SCLC).
The Cancer and Leukemia Group B 30504 trial was a randomized, placebo-controlled, phase II ...study that enrolled patients before chemotherapy (cisplatin 80 mg/m(2) or carboplatin area under the curve of 5 on day 1 plus etoposide 100 mg/m(2) per day on days 1 to 3 every 21 days for four to six cycles). Patients without progression were randomly assigned 1:1 to placebo or sunitinib 37.5 mg per day until progression. Cross-over after progression was allowed. The primary end point was progression-free survival (PFS) from random assignment for maintenance placebo versus sunitinib using a one-sided log-rank test with α = .15; 80 randomly assigned patients provided 89% power to detect a hazard ratio (HR) of 1.67.
One hundred forty-four patients were enrolled; 138 patients received chemotherapy. Ninety-five patients were randomly assigned; 10 patients did not receive maintenance therapy (five on each arm). Eighty-five patients received maintenance therapy (placebo, n = 41; sunitinib, n = 44). Grade 3 adverse events with more than 5% incidence were fatigue (19%), decreased neutrophils (14%), decreased leukocytes (7%), and decreased platelets (7%) for sunitinib and fatigue (10%) for placebo; grade 4 adverse events were GI hemorrhage (n = 1) and pancreatitis, hypocalcemia, and elevated lipase (n = 1; all in same patient) for sunitinib and thrombocytopenia (n = 1) and hypernatremia (n = 1) for placebo. Median PFS on maintenance was 2.1 months for placebo and 3.7 months for sunitinib (HR, 1.62; 70% CI, 1.27 to 2.08; 95% CI, 1.02 to 2.60; one-sided P = .02). Median overall survival from random assignment was 6.9 months for placebo and 9.0 months for sunitinib (HR, 1.28; 95% CI, 0.79 to 2.10; one-sided P = .16). Three sunitinib and no placebo patients achieved complete response during maintenance. Ten (77%) of 13 patients evaluable after cross-over had stable disease on sunitinib (6 to 27 weeks).
Maintenance sunitinib was safe and improved PFS in extensive-stage SCLC.
Background and Aims
Previous recommendations suggested living donor liver transplantation (LDLT) should not be considered for patients with Model for End‐Stage Liver Disease (MELD) > 25 and ...hepatorenal syndrome (HRS).
Approach and Results
Patients who were listed with MELD > 25 from 2008 to 2017 were analyzed with intention‐to‐treat (ITT) basis retrospectively. Patients who had a potential live donor were analyzed as ITT‐LDLT, whereas those who had none belonged to ITT‐deceased donor liver transplantation (DDLT) group. ITT‐overall survival (OS) was analyzed from the time of listing. Three hundred twenty‐five patients were listed (ITT‐LDLT n = 212, ITT‐DDLT n = 113). The risk of delist/death was lower in the ITT‐LDLT group (43.4% vs. 19.8%, P < 0.001), whereas the transplant rate was higher in the ITT‐LDLT group (78.3% vs. 52.2%, P < 0.001). The 5‐year ITT‐OS was superior in the ITT‐LDLT group (72.6% vs. 49.5%, P < 0.001) for patients with MELD > 25 and patients with both MELD > 25 and HRS (56% vs. 33.8%, P < 0.001). Waitlist mortality was the highest early after listing, and the distinct alteration of slope at survival curve showed that the benefits of ITT‐LDLT occurred within the first month after listing. Perioperative outcomes and 5‐year patient survival were comparable for patients with MELD > 25 (88% vs. 85.4%, P = 0.279) and patients with both MELD > 25 and HRS (77% vs. 76.4%, P = 0.701) after LDLT and DDLT, respectively. The LDLT group has a higher rate of renal recovery by 1 month (77.4% vs. 59.1%, P = 0.003) and 3 months (86.1% vs, 74.5%, P = 0.029), whereas the long‐term estimated glomerular filtration rate (eGFR) was similar between the 2 groups. ITT‐LDLT reduced the hazard of mortality (hazard ratio = 0.387‐0.552) across all MELD strata.
Conclusions
The ITT‐LDLT reduced waitlist mortality and allowed an earlier access to transplant. LDLT in patients with high MELD/HRS was feasible, and they had similar perioperative outcomes and better renal recovery, whereas the long‐term survival and eGFR were comparable with DDLT. LDLT should be considered for patients with high MELD/HRS, and the application of LDLT should not be restricted with a MELD cutoff.
This study explored the 10-year trajectories and outcomes of negative symptoms in patients with first-episode schizophrenia-spectrum disorder. Patients were from the historical control study ...comparing 10-year outcomes between standard care and early intervention services. A total of 298 patients were identified, 214 were successfully interviewed at 10-year follow-up for clinical and functional outcomes and 209 patients were included for final analyses. Information from clinical records were obtained systematically using standardized data entry forms. These information including negative symptoms, hospitalization and employment, monthly for year 1–3 and trimonthly for the year 4–10. Hierarchical cluster analysis was used to explore the 10-year negative symptom clusters. Demographics and early clinical characteristics related to the cluster memberships and different components of negative symptom at 10-year follow-up were further explored. The cluster analysis identified three longitudinal clusters of negative symptoms and 15% of patients were in the relapsed group. There was no difference in cluster membership between the intervention groups. Male gender and duration of hospitalizations in year four were found to be significant determinants of relapse negative symptoms. Lower education level, higher year-one negative symptom score and more months of unemployment during the first 3 years predicted overall negative symptoms at 10-year follow-up. Male gender was found to be a predictor only for avolition and anhedonia and duration of untreated psychosis only predicted anhedonia. These results highlighted the heterogeneity of longitudinal outcomes and the importance of personalized interventions.